Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
#data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')

<matplotlib.image.AxesImage at 0x23815be2048>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x23815cfecc0>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

TensorFlow Version: 1.5.0
Default GPU Device: /device:GPU:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    # TODO: Implement Function
    input_real = tf.placeholder(tf.float32, [None, image_width, image_height, image_channels], "input_real")
    input_z = tf.placeholder(tf.float32, [None, z_dim], "input_z")
    learning_rate = tf.placeholder(tf.float32, None, "learning_rate")

    return input_real, input_z, learning_rate


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the discriminator, tensor logits of the discriminator).

def discriminator(images, reuse=False):
    """
    Create the discriminator network
    :param images: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    # no alpha in args???
    alpha = 0.2
    # TODO: Implement Function
    leaky_relu = lambda x: tf.maximum(alpha * x, x)
    
    def conv(inputs, filters, batch_norm=True):
        outputs = tf.layers.conv2d(inputs, filters, 5, 2, 'same')
        if batch_norm:
            outputs = tf.layers.batch_normalization(outputs, training=True)
        return leaky_relu(outputs)
        
    
    with tf.variable_scope("discriminator", reuse=reuse):
        
        x1 = conv(images, 64, batch_norm=False)
        x2 = conv(x1, 128)
        x3 = conv(x2, 256)
        
        flat = tf.reshape(x3, (-1, 4*4*256))
        logits = tf.layers.dense(flat, 1)
        out = tf.sigmoid(logits)

        return out, logits


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    # TODO: Implement Function
    alpha = 0.2
    leaky_relu = lambda x: tf.maximum(alpha * x, x)
    with tf.variable_scope("generator", reuse=not is_train):
        x1 = tf.layers.dense(z, 7*7*512)
        x1 = tf.reshape(x1, (-1, 7, 7, 512))
        x1 = tf.layers.batch_normalization(x1, training=is_train)
        x1 = leaky_relu(x1)
        # 7*7*512
        
        x2 = tf.layers.conv2d_transpose(x1, 256, 5, 1, 'SAME')
        x2 = tf.layers.batch_normalization(x2, training=is_train)
        x2 = leaky_relu(x2)
        # 7*7*256
        
        x3 = tf.layers.conv2d_transpose(x2, 128, 5, 2, 'SAME')
        x3 = tf.layers.batch_normalization(x3, training=is_train)
        x3 = leaky_relu(x3)
        # 14*14*128
    
        logits = tf.layers.conv2d_transpose(x3, out_channel_dim, 5, 2, 'SAME')
        out = tf.tanh(logits)
        # 28*28*out_channel_dim
        return out
    return None


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

def model_loss(input_real, input_z, out_channel_dim):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    alpha = 0.9
    # TODO: Implement Function
    g_model = generator(input_z, out_channel_dim)
    d_model_real, d_logits_real = discriminator(input_real)
    d_model_fake, d_logits_fake = discriminator(g_model, reuse=True)

    d_loss_real = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real, labels=tf.ones_like(d_logits_real) * alpha))
    d_loss_fake = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.zeros_like(d_logits_fake)))
    g_loss = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.ones_like(d_logits_fake)))

    d_loss = d_loss_real + d_loss_fake

    return d_loss, g_loss

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    # TODO: Implement Function
    # Get weights and bias to update
    t_vars = tf.trainable_variables()
    d_vars = [var for var in t_vars if var.name.startswith('discriminator')]
    g_vars = [var for var in t_vars if var.name.startswith('generator')]

    # Optimize
    with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
        d_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(d_loss, var_list=d_vars)
        g_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(g_loss, var_list=g_vars)

    return d_train_opt, g_train_opt


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    # TODO: Build Model
    input_real, input_z, lr = model_inputs(data_shape[1], data_shape[2], data_shape[3], z_dim)

    d_loss, g_loss = model_loss(input_real, input_z, data_shape[3])

    d_opt, g_opt = model_opt(d_loss, g_loss, lr, beta1)
    
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count):
            steps = 0
            for batch_images in get_batches(batch_size):
                # TODO: Train Model
                steps +=1
                batch_images = batch_images * 2
                batch_z = np.random.uniform(-1, 1, size=(batch_size, z_dim))
                # Run optimizers
                _ = sess.run(d_opt, feed_dict={input_real: batch_images, input_z: batch_z, lr: learning_rate})
                _ = sess.run(g_opt, feed_dict={input_real: batch_images, input_z: batch_z, lr: learning_rate})
                
                if steps % 10 == 0:
                    train_loss_d = d_loss.eval({input_real: batch_images, input_z: batch_z})
                    train_loss_g = g_loss.eval({input_z: batch_z})

                    print("Epoch {}/{}...".format(epoch_i+1, epochs),
                          "Batch {}...".format(steps),
                          "Discriminator Loss: {:.4f}...".format(train_loss_d),
                          "Generator Loss: {:.4f}".format(train_loss_g))

                if steps % 100 == 0:
                    show_generator_output(sess, show_n_images, input_z, data_shape[3], data_image_mode)

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 32
z_dim = 100
learning_rate = 0.0002
beta1 = 0.1


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 5

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)

Epoch 1/5... Batch 10... Discriminator Loss: 5.6533... Generator Loss: 0.0062
Epoch 1/5... Batch 20... Discriminator Loss: 3.8653... Generator Loss: 0.0313
Epoch 1/5... Batch 30... Discriminator Loss: 2.9615... Generator Loss: 0.0851
Epoch 1/5... Batch 40... Discriminator Loss: 2.9858... Generator Loss: 0.0783
Epoch 1/5... Batch 50... Discriminator Loss: 2.2766... Generator Loss: 0.1674
Epoch 1/5... Batch 60... Discriminator Loss: 2.2351... Generator Loss: 0.1899
Epoch 1/5... Batch 70... Discriminator Loss: 2.1587... Generator Loss: 0.1902
Epoch 1/5... Batch 80... Discriminator Loss: 1.9944... Generator Loss: 0.2432
Epoch 1/5... Batch 90... Discriminator Loss: 2.0313... Generator Loss: 0.2431
Epoch 1/5... Batch 100... Discriminator Loss: 1.9831... Generator Loss: 0.2333

Epoch 1/5... Batch 110... Discriminator Loss: 1.6385... Generator Loss: 0.3752
Epoch 1/5... Batch 120... Discriminator Loss: 1.6900... Generator Loss: 0.3434
Epoch 1/5... Batch 130... Discriminator Loss: 1.5459... Generator Loss: 0.4467
Epoch 1/5... Batch 140... Discriminator Loss: 1.6339... Generator Loss: 0.3699
Epoch 1/5... Batch 150... Discriminator Loss: 1.6190... Generator Loss: 0.3780
Epoch 1/5... Batch 160... Discriminator Loss: 1.6620... Generator Loss: 0.3336
Epoch 1/5... Batch 170... Discriminator Loss: 1.7233... Generator Loss: 0.3166
Epoch 1/5... Batch 180... Discriminator Loss: 1.6372... Generator Loss: 0.3692
Epoch 1/5... Batch 190... Discriminator Loss: 1.4455... Generator Loss: 0.4652
Epoch 1/5... Batch 200... Discriminator Loss: 1.6858... Generator Loss: 0.3428

Epoch 1/5... Batch 210... Discriminator Loss: 1.2525... Generator Loss: 0.6684
Epoch 1/5... Batch 220... Discriminator Loss: 1.4571... Generator Loss: 0.4389
Epoch 1/5... Batch 230... Discriminator Loss: 1.5138... Generator Loss: 0.4283
Epoch 1/5... Batch 240... Discriminator Loss: 1.4465... Generator Loss: 0.4840
Epoch 1/5... Batch 250... Discriminator Loss: 1.4425... Generator Loss: 0.4753
Epoch 1/5... Batch 260... Discriminator Loss: 1.1855... Generator Loss: 1.3958
Epoch 1/5... Batch 270... Discriminator Loss: 1.2527... Generator Loss: 1.3660
Epoch 1/5... Batch 280... Discriminator Loss: 1.2021... Generator Loss: 1.0769
Epoch 1/5... Batch 290... Discriminator Loss: 1.2126... Generator Loss: 1.0558
Epoch 1/5... Batch 300... Discriminator Loss: 1.3084... Generator Loss: 1.1172

Epoch 1/5... Batch 310... Discriminator Loss: 1.2925... Generator Loss: 0.7280
Epoch 1/5... Batch 320... Discriminator Loss: 1.3716... Generator Loss: 1.2755
Epoch 1/5... Batch 330... Discriminator Loss: 1.6019... Generator Loss: 1.3783
Epoch 1/5... Batch 340... Discriminator Loss: 1.2762... Generator Loss: 1.1799
Epoch 1/5... Batch 350... Discriminator Loss: 1.3522... Generator Loss: 0.5690
Epoch 1/5... Batch 360... Discriminator Loss: 1.5840... Generator Loss: 0.3946
Epoch 1/5... Batch 370... Discriminator Loss: 1.2736... Generator Loss: 1.2756
Epoch 1/5... Batch 380... Discriminator Loss: 1.1577... Generator Loss: 0.9530
Epoch 1/5... Batch 390... Discriminator Loss: 1.0833... Generator Loss: 1.0217
Epoch 1/5... Batch 400... Discriminator Loss: 1.1722... Generator Loss: 1.0542

Epoch 1/5... Batch 410... Discriminator Loss: 1.4410... Generator Loss: 0.5318
Epoch 1/5... Batch 420... Discriminator Loss: 1.7594... Generator Loss: 0.3453
Epoch 1/5... Batch 430... Discriminator Loss: 1.3300... Generator Loss: 0.6037
Epoch 1/5... Batch 440... Discriminator Loss: 1.7237... Generator Loss: 0.3236
Epoch 1/5... Batch 450... Discriminator Loss: 1.4654... Generator Loss: 0.4494
Epoch 1/5... Batch 460... Discriminator Loss: 1.2023... Generator Loss: 1.3904
Epoch 1/5... Batch 470... Discriminator Loss: 1.1726... Generator Loss: 0.7552
Epoch 1/5... Batch 480... Discriminator Loss: 1.6334... Generator Loss: 0.3442
Epoch 1/5... Batch 490... Discriminator Loss: 1.6352... Generator Loss: 0.3642
Epoch 1/5... Batch 500... Discriminator Loss: 1.4091... Generator Loss: 0.4595

Epoch 1/5... Batch 510... Discriminator Loss: 1.3656... Generator Loss: 0.6607
Epoch 1/5... Batch 520... Discriminator Loss: 1.0578... Generator Loss: 0.8537
Epoch 1/5... Batch 530... Discriminator Loss: 1.0701... Generator Loss: 1.2101
Epoch 1/5... Batch 540... Discriminator Loss: 1.5196... Generator Loss: 1.7573
Epoch 1/5... Batch 550... Discriminator Loss: 1.1144... Generator Loss: 0.9889
Epoch 1/5... Batch 560... Discriminator Loss: 1.2137... Generator Loss: 1.4808
Epoch 1/5... Batch 570... Discriminator Loss: 1.4065... Generator Loss: 1.3041
Epoch 1/5... Batch 580... Discriminator Loss: 1.1770... Generator Loss: 1.0516
Epoch 1/5... Batch 590... Discriminator Loss: 1.0785... Generator Loss: 1.2036
Epoch 1/5... Batch 600... Discriminator Loss: 1.5566... Generator Loss: 0.3973

Epoch 1/5... Batch 610... Discriminator Loss: 1.1738... Generator Loss: 0.8417
Epoch 1/5... Batch 620... Discriminator Loss: 1.3840... Generator Loss: 0.5469
Epoch 1/5... Batch 630... Discriminator Loss: 1.5069... Generator Loss: 0.4303
Epoch 1/5... Batch 640... Discriminator Loss: 1.2334... Generator Loss: 0.9725
Epoch 1/5... Batch 650... Discriminator Loss: 1.1227... Generator Loss: 1.0563
Epoch 1/5... Batch 660... Discriminator Loss: 1.5960... Generator Loss: 0.3875
Epoch 1/5... Batch 670... Discriminator Loss: 1.3340... Generator Loss: 0.8848
Epoch 1/5... Batch 680... Discriminator Loss: 1.6705... Generator Loss: 0.3354
Epoch 1/5... Batch 690... Discriminator Loss: 1.5504... Generator Loss: 0.3887
Epoch 1/5... Batch 700... Discriminator Loss: 1.2528... Generator Loss: 0.6131

Epoch 1/5... Batch 710... Discriminator Loss: 1.5578... Generator Loss: 0.4067
Epoch 1/5... Batch 720... Discriminator Loss: 1.5850... Generator Loss: 0.3655
Epoch 1/5... Batch 730... Discriminator Loss: 1.3343... Generator Loss: 0.5222
Epoch 1/5... Batch 740... Discriminator Loss: 1.4145... Generator Loss: 0.4964
Epoch 1/5... Batch 750... Discriminator Loss: 1.2446... Generator Loss: 0.6339
Epoch 1/5... Batch 760... Discriminator Loss: 1.4253... Generator Loss: 2.1071
Epoch 1/5... Batch 770... Discriminator Loss: 1.1055... Generator Loss: 1.1953
Epoch 1/5... Batch 780... Discriminator Loss: 1.1975... Generator Loss: 1.0332
Epoch 1/5... Batch 790... Discriminator Loss: 1.4295... Generator Loss: 0.4950
Epoch 1/5... Batch 800... Discriminator Loss: 1.1665... Generator Loss: 1.0490

Epoch 1/5... Batch 810... Discriminator Loss: 1.0775... Generator Loss: 0.8654
Epoch 1/5... Batch 820... Discriminator Loss: 1.2763... Generator Loss: 0.5919
Epoch 1/5... Batch 830... Discriminator Loss: 2.0284... Generator Loss: 0.2789
Epoch 1/5... Batch 840... Discriminator Loss: 1.1076... Generator Loss: 1.0017
Epoch 1/5... Batch 850... Discriminator Loss: 1.5346... Generator Loss: 0.4107
Epoch 1/5... Batch 860... Discriminator Loss: 1.1972... Generator Loss: 0.6865
Epoch 1/5... Batch 870... Discriminator Loss: 1.2327... Generator Loss: 1.4840
Epoch 1/5... Batch 880... Discriminator Loss: 1.2027... Generator Loss: 0.7964
Epoch 1/5... Batch 890... Discriminator Loss: 1.1783... Generator Loss: 1.5230
Epoch 1/5... Batch 900... Discriminator Loss: 1.1628... Generator Loss: 0.7065

Epoch 1/5... Batch 910... Discriminator Loss: 1.3040... Generator Loss: 0.5600
Epoch 1/5... Batch 920... Discriminator Loss: 2.5129... Generator Loss: 0.1690
Epoch 1/5... Batch 930... Discriminator Loss: 1.2987... Generator Loss: 0.5819
Epoch 1/5... Batch 940... Discriminator Loss: 1.4360... Generator Loss: 0.4788
Epoch 1/5... Batch 950... Discriminator Loss: 1.1482... Generator Loss: 0.7468
Epoch 1/5... Batch 960... Discriminator Loss: 0.9836... Generator Loss: 1.0382
Epoch 1/5... Batch 970... Discriminator Loss: 1.1132... Generator Loss: 0.8279
Epoch 1/5... Batch 980... Discriminator Loss: 1.0114... Generator Loss: 1.3444
Epoch 1/5... Batch 990... Discriminator Loss: 1.2354... Generator Loss: 0.7852
Epoch 1/5... Batch 1000... Discriminator Loss: 1.1550... Generator Loss: 0.7259

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

batch_size = 32
z_dim = 100
learning_rate = 0.0002
beta1 = 0.1


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode)

Epoch 1/2... Batch 10... Discriminator Loss: 4.9688... Generator Loss: 0.0149
Epoch 1/2... Batch 20... Discriminator Loss: 5.6123... Generator Loss: 0.0074
Epoch 1/2... Batch 30... Discriminator Loss: 3.1060... Generator Loss: 0.0828
Epoch 1/2... Batch 40... Discriminator Loss: 2.4479... Generator Loss: 0.1956
Epoch 1/2... Batch 50... Discriminator Loss: 1.6389... Generator Loss: 0.4734
Epoch 1/2... Batch 60... Discriminator Loss: 1.6807... Generator Loss: 1.2218
Epoch 1/2... Batch 70... Discriminator Loss: 1.5885... Generator Loss: 1.1081
Epoch 1/2... Batch 80... Discriminator Loss: 1.7634... Generator Loss: 1.0323
Epoch 1/2... Batch 90... Discriminator Loss: 1.7413... Generator Loss: 0.8534
Epoch 1/2... Batch 100... Discriminator Loss: 2.2495... Generator Loss: 0.1760

Epoch 1/2... Batch 110... Discriminator Loss: 2.0090... Generator Loss: 0.2450
Epoch 1/2... Batch 120... Discriminator Loss: 1.9015... Generator Loss: 0.2585
Epoch 1/2... Batch 130... Discriminator Loss: 1.7732... Generator Loss: 0.4197
Epoch 1/2... Batch 140... Discriminator Loss: 1.8760... Generator Loss: 1.5388
Epoch 1/2... Batch 150... Discriminator Loss: 1.4628... Generator Loss: 0.8939
Epoch 1/2... Batch 160... Discriminator Loss: 1.9761... Generator Loss: 0.2350
Epoch 1/2... Batch 170... Discriminator Loss: 1.9589... Generator Loss: 0.3395
Epoch 1/2... Batch 180... Discriminator Loss: 2.1668... Generator Loss: 0.2153
Epoch 1/2... Batch 190... Discriminator Loss: 1.5068... Generator Loss: 0.6701
Epoch 1/2... Batch 200... Discriminator Loss: 1.4018... Generator Loss: 1.1578

Epoch 1/2... Batch 210... Discriminator Loss: 1.2965... Generator Loss: 0.7136
Epoch 1/2... Batch 220... Discriminator Loss: 1.4975... Generator Loss: 0.4247
Epoch 1/2... Batch 230... Discriminator Loss: 2.2740... Generator Loss: 0.1752
Epoch 1/2... Batch 240... Discriminator Loss: 1.5926... Generator Loss: 0.8391
Epoch 1/2... Batch 250... Discriminator Loss: 1.3248... Generator Loss: 1.7143
Epoch 1/2... Batch 260... Discriminator Loss: 1.3954... Generator Loss: 1.1871
Epoch 1/2... Batch 270... Discriminator Loss: 1.6450... Generator Loss: 1.3027
Epoch 1/2... Batch 280... Discriminator Loss: 1.5408... Generator Loss: 1.3196
Epoch 1/2... Batch 290... Discriminator Loss: 1.5542... Generator Loss: 0.5545
Epoch 1/2... Batch 300... Discriminator Loss: 1.6442... Generator Loss: 0.7698

Epoch 1/2... Batch 310... Discriminator Loss: 1.6163... Generator Loss: 0.4553
Epoch 1/2... Batch 320... Discriminator Loss: 1.7701... Generator Loss: 0.3197
Epoch 1/2... Batch 330... Discriminator Loss: 1.5829... Generator Loss: 0.5430
Epoch 1/2... Batch 340... Discriminator Loss: 1.5302... Generator Loss: 0.4975
Epoch 1/2... Batch 350... Discriminator Loss: 1.4370... Generator Loss: 0.5149
Epoch 1/2... Batch 360... Discriminator Loss: 1.5814... Generator Loss: 0.4335
Epoch 1/2... Batch 370... Discriminator Loss: 1.6202... Generator Loss: 0.4280
Epoch 1/2... Batch 380... Discriminator Loss: 1.6596... Generator Loss: 0.3667
Epoch 1/2... Batch 390... Discriminator Loss: 1.3399... Generator Loss: 0.8689
Epoch 1/2... Batch 400... Discriminator Loss: 1.3569... Generator Loss: 1.1003

Epoch 1/2... Batch 410... Discriminator Loss: 1.4213... Generator Loss: 0.5130
Epoch 1/2... Batch 420... Discriminator Loss: 1.7484... Generator Loss: 0.3234
Epoch 1/2... Batch 430... Discriminator Loss: 1.5780... Generator Loss: 0.4421
Epoch 1/2... Batch 440... Discriminator Loss: 1.2844... Generator Loss: 1.1436
Epoch 1/2... Batch 450... Discriminator Loss: 1.3149... Generator Loss: 0.9613
Epoch 1/2... Batch 460... Discriminator Loss: 1.5986... Generator Loss: 0.4002
Epoch 1/2... Batch 470... Discriminator Loss: 1.9749... Generator Loss: 0.2846
Epoch 1/2... Batch 480... Discriminator Loss: 1.5733... Generator Loss: 0.4557
Epoch 1/2... Batch 490... Discriminator Loss: 1.3212... Generator Loss: 0.9938
Epoch 1/2... Batch 500... Discriminator Loss: 1.9528... Generator Loss: 0.2503

Epoch 1/2... Batch 510... Discriminator Loss: 1.3977... Generator Loss: 0.5564
Epoch 1/2... Batch 520... Discriminator Loss: 1.2610... Generator Loss: 1.1234
Epoch 1/2... Batch 530... Discriminator Loss: 1.5183... Generator Loss: 0.7819
Epoch 1/2... Batch 540... Discriminator Loss: 1.1301... Generator Loss: 1.1210
Epoch 1/2... Batch 550... Discriminator Loss: 1.4492... Generator Loss: 0.8471
Epoch 1/2... Batch 560... Discriminator Loss: 1.3044... Generator Loss: 1.7408
Epoch 1/2... Batch 570... Discriminator Loss: 1.1688... Generator Loss: 1.3898
Epoch 1/2... Batch 580... Discriminator Loss: 1.5199... Generator Loss: 0.7569
Epoch 1/2... Batch 590... Discriminator Loss: 1.9449... Generator Loss: 1.9607
Epoch 1/2... Batch 600... Discriminator Loss: 1.0131... Generator Loss: 1.7976

Epoch 1/2... Batch 610... Discriminator Loss: 0.8703... Generator Loss: 1.4087
Epoch 1/2... Batch 620... Discriminator Loss: 1.3270... Generator Loss: 1.5564
Epoch 1/2... Batch 630... Discriminator Loss: 1.4061... Generator Loss: 1.3606
Epoch 1/2... Batch 640... Discriminator Loss: 1.2586... Generator Loss: 0.6188
Epoch 1/2... Batch 650... Discriminator Loss: 1.2881... Generator Loss: 1.2718
Epoch 1/2... Batch 660... Discriminator Loss: 1.0623... Generator Loss: 1.4423
Epoch 1/2... Batch 670... Discriminator Loss: 1.6688... Generator Loss: 1.0521
Epoch 1/2... Batch 680... Discriminator Loss: 1.1327... Generator Loss: 1.4349
Epoch 1/2... Batch 690... Discriminator Loss: 1.2098... Generator Loss: 1.8257
Epoch 1/2... Batch 700... Discriminator Loss: 1.1277... Generator Loss: 1.6647

Epoch 1/2... Batch 710... Discriminator Loss: 0.8229... Generator Loss: 2.1453
Epoch 1/2... Batch 720... Discriminator Loss: 0.9013... Generator Loss: 1.1667
Epoch 1/2... Batch 730... Discriminator Loss: 1.3797... Generator Loss: 1.2350
Epoch 1/2... Batch 740... Discriminator Loss: 1.3780... Generator Loss: 0.4949
Epoch 1/2... Batch 750... Discriminator Loss: 1.2540... Generator Loss: 0.5514
Epoch 1/2... Batch 760... Discriminator Loss: 1.7165... Generator Loss: 0.3586
Epoch 1/2... Batch 770... Discriminator Loss: 1.4531... Generator Loss: 0.4495
Epoch 1/2... Batch 780... Discriminator Loss: 1.2786... Generator Loss: 0.5329
Epoch 1/2... Batch 790... Discriminator Loss: 0.6676... Generator Loss: 1.5933
Epoch 1/2... Batch 800... Discriminator Loss: 1.1439... Generator Loss: 0.7292

Epoch 1/2... Batch 810... Discriminator Loss: 1.3545... Generator Loss: 0.4932
Epoch 1/2... Batch 820... Discriminator Loss: 0.8983... Generator Loss: 0.9335
Epoch 1/2... Batch 830... Discriminator Loss: 1.1653... Generator Loss: 4.1377
Epoch 1/2... Batch 840... Discriminator Loss: 0.6066... Generator Loss: 1.6298
Epoch 1/2... Batch 850... Discriminator Loss: 0.8323... Generator Loss: 3.3956
Epoch 1/2... Batch 860... Discriminator Loss: 0.6802... Generator Loss: 1.4249
Epoch 1/2... Batch 870... Discriminator Loss: 0.9087... Generator Loss: 0.9274
Epoch 1/2... Batch 880... Discriminator Loss: 0.8240... Generator Loss: 0.9938
Epoch 1/2... Batch 890... Discriminator Loss: 0.5803... Generator Loss: 1.7268
Epoch 1/2... Batch 900... Discriminator Loss: 0.7106... Generator Loss: 1.3588

Epoch 1/2... Batch 910... Discriminator Loss: 1.3487... Generator Loss: 4.0272
Epoch 1/2... Batch 920... Discriminator Loss: 0.7600... Generator Loss: 1.1462
Epoch 1/2... Batch 930... Discriminator Loss: 0.7432... Generator Loss: 3.5183
Epoch 1/2... Batch 940... Discriminator Loss: 0.4619... Generator Loss: 2.4710
Epoch 1/2... Batch 950... Discriminator Loss: 0.4761... Generator Loss: 3.4845
Epoch 1/2... Batch 960... Discriminator Loss: 0.7934... Generator Loss: 1.2129
Epoch 1/2... Batch 970... Discriminator Loss: 0.6208... Generator Loss: 1.4532
Epoch 1/2... Batch 980... Discriminator Loss: 1.5419... Generator Loss: 3.8899
Epoch 1/2... Batch 990... Discriminator Loss: 0.9759... Generator Loss: 1.3178
Epoch 1/2... Batch 1000... Discriminator Loss: 0.6769... Generator Loss: 1.4925

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.